Abstract
Here we demonstrate a spectrally broadband, gigahertz-fast Mach–Zehnder interferometric modulator exhibiting a miniscule ${{\rm V}_\pi }{\rm L}$ of $95\;{\rm V} \cdot \unicode{x00B5}{\rm m}$, deploying a subwavelength short electrostatically tunable plasmonic phase shifter based on heterogeneously integrated indium tin oxide thin films into silicon photonics.
© 2020 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
Indium tin oxide (ITO), belonging to the class of transparent conductive oxides, is a material extensively adopted in high-tech industry such as in touchscreen displays of smartphones or contacts for solar cells. Recently, ITO has been explored for electro-optic modulation using its free-carrier dispersive effect enabling unity-strong index modulation [1–3]. However, gigahertz (GHz)-fast modulation capability using ITO is yet to be demonstrated—a feature we show herein. Given the ubiquitous usage of phase-shifter (PS) technologies, such as in data communication, optical phased arrays, analog and RF photonics, sensing, and so on, here we focus on a Mach–Zehnder interferometer (MZI)-based modulator to demonstrate a comprehensive platform of heterogeneous integration of ITO-based opto-electronics into silicon photonic integrated circuits (PIC). Since the real part of the optical refractive index ($n$) is of interest in PSs, in previous studies we have shown the interplay between a selected optical mode (e.g., photonic bulk versus plasmonic) and the material’s figure of merit ($\Delta n/\Delta \alpha $), where $\alpha $ is the optical loss, directly resultant from Kramers–Kronig relations [4]. Additionally, ITO can be selectively prepared (via process conditions [5]) for operating in either an $n$-dominant or $\alpha $-dominated region [4], demonstrating a photonic-mode ITO-oxide-Si MZI on silicon photonics with an efficient ${{\rm V}_\pi }{\rm L}={0.52}\;{\rm V} \cdot {\rm mm}$ [2] and a plasmonic version deploying a lateral gate exhibiting a ${{\rm V}_\pi }{\rm L}={0.063}\;{\rm V}\cdot {\rm mm}$ [6]. Indeed, a plasmonic mode enables a strong light–matter interaction (e.g., extrinsic slow-light effect), which, when superimposed with ITO’s intrinsic slow-light effect, proximal epsilon-near-zero (ENZ) effects [7], enables realization of just 1–3 µm short PSs [4], allowing small (${\sim}{\rm fF}$) electrical capacitances for efficient and fast signal modulation. Here we design the ITO material parameters to control operation in the $n$-dominant region adequately close to but not at the high-loss ENZ (ENZ located in the $\alpha $-dominant region) [4]. In fact, unlike lithium niobate (LN) optoelectronics requiring careful crystal-orientation control [8,9], ITO thin films are crystal-orientation independent and feature intrinsically uniform optical characteristics as deposited. Here we discuss an ITO-plasmon-based PS heterogeneously integrated into a silicon photonic MZI delivering GHz-fast broadband modulation and thus open opportunities for multispectral operation.
The base interferometer is taped out as a symmetric silicon-on-insulator (SOI) MZI to minimize chirp effects induced by different splitting ratios in the Y junctions of the MZI and includes post-tape out loss balancing between both arms using a metallic strip (${L_b}$) on the nonmodulated arm to minimize extinction ratio (ER) degradation [Fig. 1(a)]. Sweep of the active PS device length (${L_d}$) ranges from sub-$\lambda $ (1.4 µm) to $\lambda $-scale devices (3.5 µm) [Fig. 1(b)]. Broadband spectral response is measured in the C band [${\sim}{30}\;{\rm nm}$, Fig. 2(a)], which is expected since the plasmonic resonance of the mode has a FWHM ${\sim}{100{\rm s}}$ of nanometers (nm). The spectral response is determined by ITO dispersion and proximity to the ENZ. For ultrabroadband applications (e.g., ${100 + }\;{\rm nm}$) ITO modulators for different spectral regions (e.g., $\Delta \lambda ={50}\;{\rm nm}$) can be processed using different conditions [5]. Functional capacitor traits in the measured bias range are observed [Fig. 2(b)]. DC electro-optic transmission power tests and squared cosine fit (as dictated by MZI operating principle) result in an ER of ${\sim}{3}$ to $ \gt {8}{\rm dB}$, respectively [Fig. 2(c)]. The measured ${{\rm V}_\pi }{\rm L}$ is just $95 \pm 2\;{\rm V} \cdot \unicode{x00B5}{\rm m}$ and rather constant across all device scaling. The results indicate a modal index change $\Delta {n_{\rm eff}}$ of ${\sim}{0.2}$ [Fig. 2(d)], and FEM eigenmode analysis [inset, Fig. 1(b)] reveals an ITO index change of about 0.6 [Fig. 2(e)] reflecting an ${\sim}{2} \times $ increased confinement factor ($\Gamma $) corresponding to active biasing, slightly lower than previous ITO modulators [2], and intentionally enabling lower insertion loss (IL) of about 6.7 dB. Cutback measurements reveal 1.6 dB/µm propagation loss in the active region and an additional 1.3 dB/coupling loss from in/out coupling of the mode from the Si waveguide, while the passive loss balancing contact [Fig. 1(a), ${L_b}$] exhibits a 1.2 dB/µm propagation loss and 1.1 dB/coupling loss, correspondingly. Note that the high loss per unit length in plasmonics is alleviated by an enhanced light–matter interaction enabling $\lambda $-short device lengths (${L_d}$); thus the total IL is comparable to Si photonic MZIs. The deposited ITO thin film carrier concentration ${N_c}$ of ${3.1} \times {{10}^{20}}\;{{\rm cm}^{ - 3}}$ is determined from metrology, and a change $\Delta {N_c}={2.1} \times {{10}^{20}}\;{{\rm cm}^{ - 3}}$ estimated from the gated measurements suggests $n$-dominant operation, however intentionally away from the high-loss ENZ (${6 - 7} \times {{10}^{20}}\;{{\rm cm}^{ - 3}}$) state, yet sufficiently near to capture a slow-light effect [4].
Frequency response (${{\rm S}_{21}}$) is obtained by generating a low power modulating signal (0 dBm) with a 50 GHz network analyzer; a bias tee combines DC voltage (6 V) with the RF signal [Fig. 2(f)]. RF output from the modulator is amplified using a broadband erbium-doped fiber amplifier (EDFA, ${\sim}{35}\;{\rm dB}$), and an optical tunable filter reduces undesired noise by 20 dB. The modulated light is collected by a photodetector. The ${-}{3}\;{\rm dB}$ roll-off (small signal) shows a speed of 1.1 GHz [Fig. 2(g)], which matches estimations for the RC delay given capacitance of 213 fF and total resistance of 680 Ω, while dynamic switching energy (${\sim}{\rm pJ}$) characterizes the spectral trade-off [2]. Performance comparison of this ITO paradigm with recent modulators shows similar achievable speeds, allowing for CMOS low drive voltages and competent ${{\rm V}_\pi }{\rm L}$ enabled by efficient electrostatics (${t_{\rm ox}}={5}\;{\rm nm}$, ${{\rm Hf}_3}{{\rm N}_4}$, pad-overlay optimization, annealing, plasma treatment), which is fundamentally challenging in LN due to its delicate loss sensitivity (Table 1).
This GHz-fast broadband integrated modulator bears relevance since ITO is a foundry-compatible material. Unlike the crystal-orientation-sensitive LN, ITO optoelectronics is synergistic to enhancing electrostatics known from transistor technology.
Funding
Air Force Office of Scientific Research (FA9550-17-1-0071, FA9550-17-1-0377).
Disclosures
The authors declare no conflicts of interest.
REFERENCES
1. V. J. Sorger, N. D. Lanzillotti- Kimura, R. Ma, and X. Zhang, Nanophotonics 1, 17 (2012). [CrossRef]
2. R. Amin, R. Maiti, C. Carfano, Z. Ma, M. H. Tahersima, Y. Lilach, D. Ratnayake, H. Dalir, and V. J. Sorger, APL Photon. 3, 126104 (2018). [CrossRef]
3. E. Li, Q. Gao, S. Liverman, and A. X. Wang, Opt. Lett. 43, 4429 (2018). [CrossRef]
4. R. Amin, C. Suer, Z. Ma, I. Sarpkaya, J. B. Khurgin, R. Agarwal, and V. J. Sorger, Nanophotonics 7, 455 (2017). [CrossRef]
5. Y. Gui, M. Miscuglio, Z. Ma, M. H. Tahersima, S. Sun, R. Amin, H. Dalir, and V. J. Sorger, Sci. Rep. 9, 11279 (2019). [CrossRef]
6. R. Amin, R. Maiti, J. K. George, X. Ma, Z. Ma, H. Dalir, M. Miscuglio, and V. J. Sorger, J. Lightwave Technol. 38, 282 (2020). [CrossRef]
7. N. Kinsey and J. B. Khurgin, Opt. Mater. Express 9, 2793 (2019). [CrossRef]
8. C. Wang, M. Zhang, X. Chen, M. Bertrand, A. Shams-Ansari, S. Chandrasekhar, P. Winzer, and M. Lončar, Nature 562, 101 (2018). [CrossRef]
9. M. Thomaschewski, V. A. Zenin, C. Wolff, and S. I. Bozhevolnyi, Nat. Commun. 11, 748 (2020). [CrossRef]
10. M. Ayata, Y. Fedoryshyn, W. Heni, B. Baeuerle, A. Josten, M. Zahner, U. Koch, Y. Salamin, C. Hoessbacher, C. Haffner, D. L. Elder, L. R. Dalton, and J. Leuthold, Science 358, 630 (2017). [CrossRef]
11. S. Abel, F. Eltes, J. E. Ortmann, A. Messner, P. Castera, T. Wagner, D. Urbonas, A. Rosa, A. M. Gutierrez, D. Tulli, P. Ma, B. Baeuerle, A. Josten, W. Heni, D. Caimi, L. Czornomaz, A. A. Demkov, J. Leuthold, P. Sanchis, and J. Fompeyrine, Nat. Mater. 18, 42 (2019). [CrossRef]
12. C. Haffner, D. Chelladurai, Y. Fedoryshyn, A. Josten, B. Baeuerle, W. Heni, T. Watanabe, T. Cui, B. Cheng, S. Saha, D. L. Elder, L. R. Dalton, A. Boltasseva, V. M. Shalaev, N. Kinsey, and J. Leuthold, Nature 556, 483 (2018). [CrossRef]